Author Topic: 5 Gsample/s, 1 GHz, 4 ch scope advise  (Read 12715 times)

0 Members and 1 Guest are viewing this topic.

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #25 on: November 28, 2015, 12:46:23 pm »
... comes with Danaher grade non-support.

That's a bit harsh, I'm sure they'll be working on it:   ;)

 

Offline robert_

  • Regular Contributor
  • *
  • Posts: 151
  • Country: de
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #26 on: November 28, 2015, 01:02:26 pm »
As for the RTE, we had an offer (for 350Mhz) that was well below 7k, so its quite cheap, just 1k extra on the RTM. And a good 3.5k below a HDO6k.
Sadly, as always, bosses decide only on numbers, guess why i have a 100Mhz DPO3k on my bench every day. Its f.ing cheap but well it DOES suck.
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #27 on: November 28, 2015, 01:30:25 pm »
As for the RTE, we had an offer (for 350Mhz) that was well below 7k, so its quite cheap, just 1k extra on the RTM.

EUR or US$? But I guess after taxes the EUR amount comes out at roughly the same as the US$ figure anyways.

Depending on how much it is below $7k (i.e. the 4Ch 350MHz WaveSurfer 3034 costs around $5.5k without any discounts, just as a comparison) the price might be good, but don't forget that you're still stuck with the interface and other niggles. Depending on what you need, you might even be better off with an RTM (which has a simpler and more intuitive interface, plus can have the spectrum option).

Quote
And a good 3.5k below a HDO6k.
Sadly, as always, bosses decide only on numbers, guess why i have a 100Mhz DPO3k on my bench every day. Its f.ing cheap but well it DOES suck.

Well, you could try to remind your boss that your time costs money as well, and if you have to fight your tool then any savings on purchase can quickly be offset by time wasted.
 

Online KE5FX

  • Super Contributor
  • ***
  • Posts: 1111
  • Country: us
    • KE5FX.COM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #28 on: November 28, 2015, 10:24:32 pm »
In the real world, 2 MS is a lot of sample data, 64K is a long FFT record, and sub-millivolt signals are often best observed with said FFT.

Not in 2015, no, and even less so for a 1GHz+ scope with sample rates in excess of 4GSa/s. 2Mpts memory was pretty good (not stellar, though) 15 years ago, however even back then 64k FFT was pretty poor.

Curious, what do you tend to work on (either at home or work) that frequently needs longer sample records and larger FFT kernels?

Quote
To offer a scope like the Keysight DSOX4k in the price range it sits in with 4M memory and 64k FFT is a pretty bad joke in this day and age.

How the memory is used is much more important than how much there is, in my experience.  Everyone's application is different, but from what I've seen in the scope business, it is far too easy to get caught up in "bench racing" and specsmanship.  You end up with a crappier instrument, albeit one with larger numbers on the sell sheet. 

It's much like the thread on microscopes, where Rupunzell points out that quality can't really be inferred from the specs alone.  I don't care that much about my microscope, but I rely on my oscilloscope to cover my own shortcomings as an engineer, so it's almost literally a part of me.  :)

Parameters like record length aren't limited only by marketing considerations in the Agilent/Keysight scopes, the way they are in everyone else's.  They are limited by the ASICs that they use to process and display data.  The sample record is shorter in my MSO6000-series scope than it would be in an optioned-out Tek MDO, but the fact that I don't have to fiddle with it to get the display I want counts for a lot.  Let's see, which button do I press now?  Do I want a "fast acquisition" or a "long acquisition"?  Gee, I dunno.  How about you just show me the signal?

Quote
I guess you haven't had to use their high end scopes (i.e. the DSO90k) where some firmware versions suffered from jumping encoders, wrong indications (i.e. showing channels as set to GND when they weren't) and other silly things. Granted, these problems were have been fixed, but Agilent didn't exactly rush out to squash these bugs.

This is true.  I've been lucky so far, in that I've been able to stick with the last generation of Agilent scopes to run VxWorks.  The controls are responsive and reliable (if not the best-feeling.)  Not many firmware bugs, either.  I don't look forward to "upgrading" to a scope running a heavyweight desktop OS...
« Last Edit: November 28, 2015, 10:28:27 pm by KE5FX »
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #29 on: November 29, 2015, 11:32:25 am »
Curious, what do you tend to work on (either at home or work) that frequently needs longer sample records and larger FFT kernels?

For example, I'm doing a lot of signal analysis of complex non-standard communication signals, which probably isn't a common use case.

But that's besides the point anyways, as larger sample memory is useful in lots of standard cases as well, as it means the scope will maintain a high sample rate at much longer timebase settings.

As for FFT, the number of points used decides the frequency resolution (RBW), which is inversely proportional to the sample length.

Quote
Quote
To offer a scope like the Keysight DSOX4k in the price range it sits in with 4M memory and 64k FFT is a pretty bad joke in this day and age.

How the memory is used is much more important than how much there is, in my experience. 

Sometimes you can compensate or work around sample memory limitations to an extend, but at the end of the day if the sample memory is small then this will limit a scope's performance.

Also, don't forget that we're not talking about some bottom-of-the-barrel scope or some old 2nd Tek scope in this thread, we're talking about modern mid-range scopes priced between some $7k to over $20k. Heck, even a $3k WaveSurfer 3022 offers 10Mpts sample and 1M FFT.

Quote
Everyone's application is different, but from what I've seen in the scope business, it is far too easy to get caught up in "bench racing" and specsmanship.  You end up with a crappier instrument, albeit one with larger numbers on the sell sheet. 

That is true to some extend (i.e. the larger Rigol scopes which have lots of sample memory but otherwise perform poorly), but generally dismissing progress made over the years is not the right answer.

The simple point is that Keysight's DSOX4k offers *today* certain specs like sample memory and FFT that were considered good for a scope in its class one and a half decade ago, at the same price or even above the price other competitors ask for while offering better specifications (and the small sample memory or the 64k FFT are not the only limitations of the DSOX4k).

Quote
It's much like the thread on microscopes, where Rupunzell points out that quality can't really be inferred from the specs alone.

I certainly don't want to go deeper into what someone who regularly refers to analog scopes as "high fidelity" instruments stated but the validity of that statement pretty much depends on your definition of "quality" (the meaning of which can vary a lot).

At the end of the day, what scope specs give you is a set of parameters and limitations the manufacturer warrants its instrument to achieve in the defined environment. So in terms of "quality" of capturing and processing signals, the specs actually tell you quite a lot about how a scope will perform. It's up to you to weight them against each other in regards to your specific use case.

What specs don't tell you of course is how well the scope is built, how cumbersome the UI is, how the decoders "feel" and stuff like that. But a lot of that is subjective anyways, and not something that can be stated in the specs.

That's why for any expensive test equipment purchase one should always try to get a loaner of the instruments that are considered for purchase.

Quote
I don't care that much about my microscope, but I rely on my oscilloscope to cover my own shortcomings as an engineer, so it's almost literally a part of me.  :)

Not sure if it's shortcomings or just a bit of lack of self confidence  ;) but I agree, a scope is one of the most important instruments for many EEs. Which is why doing the leg work and researching properly about all available alternatives before buying is important.

Quote
Parameters like record length aren't limited only by marketing considerations in the Agilent/Keysight scopes, the way they are in everyone else's.  They are limited by the ASICs that they use to process and display data. 

I know, and Keysight have painted themselves into a corner here, but at the end of the day their scopes compete on a market with other scopes, and in this day and age there are enough better alternatives on the market so they have to up their game by quite some margin to become competitive. But I guess they'll try to milk the "Tek or HP/Agilent/Keysight only" crowds for a while longer. If that works for them, fair enough.

Quote
The sample record is shorter in my MSO6000-series scope than it would be in an optioned-out Tek MDO, but the fact that I don't have to fiddle with it to get the display I want counts for a lot.  Let's see, which button do I press now?  Do I want a "fast acquisition" or a "long acquisition"?  Gee, I dunno.  How about you just show me the signal?

I agree, but that's not because of the longer memory but simply because the Tek's UI sucks. Other scopes manage memory automatically (which can be overridden by the user).

Quote
Quote
I guess you haven't had to use their high end scopes (i.e. the DSO90k) where some firmware versions suffered from jumping encoders, wrong indications (i.e. showing channels as set to GND when they weren't) and other silly things. Granted, these problems were have been fixed, but Agilent didn't exactly rush out to squash these bugs.

This is true.  I've been lucky so far, in that I've been able to stick with the last generation of Agilent scopes to run VxWorks.  The controls are responsive and reliable (if not the best-feeling.)  Not many firmware bugs, either.

Yes, today. And that is only because they've seen a large number of firmware updates over the years to squash the many bugs these scopes came to market with. They weren't released in that state.

Don't get me wrong, I do like these VxWorks based scopes, once the major bugs were fixed they were great scopes at their time. I also liked the VxWorks based LeCroy WaveRunner2/WavePro 900 scopes, which could do a lot of advanced stuff that even today only few high end scopes can do.

But would I pay $7k (or whatever their original price was back then) for a new scope with the same specs today? Hell, no!

Quote
I don't look forward to "upgrading" to a scope running a heavyweight desktop OS...

I don't know, I really like Windows scopes. They take a bit longer to boot obviously but having access to analysis tools directly on your instrument is great, plus there are all the small benefits like not being limited to FAT formatted USB sticks (being able to use USB hard drives or exFAT formatted sticks and memory cards is nice).

Truly embedded OSes like VxWorks and such are great for low-end and entry-level scopes but for a modern upper mid-range or high-end scope I'd not want anything else than a full Windows scope.
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 10694
  • Country: gb
    • Having fun doing more, with less
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #30 on: November 29, 2015, 11:56:43 am »
Sometimes you can compensate or work around sample memory limitations to an extend, but at the end of the day if the sample memory is small then this will limit a scope's performance.

OK, convince me for a "typical case" user of an oscilloscope looking at the time-domain representation of a signal. Oscilloscope != spectrum/vector/network analyser.

If you have a long duration signal, when isn't is sufficient to digitise at the full rate? Provided you then present three values at each displayed point: mean, min and max so as to attempt to mimic the delightful informative "fuzziness" that can be seen on analogue scopes.

Background: I've been using analogue scopes for 40 years, and digital scopes for 25 years. I have my own opinions as to the advantages, disadvantages, and overlap of each. I like analogue displays, and eye diagrams (and similar "heat maps") when appropriate. I once (in ~1992) had a student build a deep digitiser for just your use case, analysis of comms signals. I'm a strong believer of debugging digital signals in the digital domain, and that almost all signals are analogue (except where we choose to interpret them as digital).

Hence I will claim that anything equivalent to using a scope to capturing and decoding, for example, the data on an serial peripheral interface (SPI) is misguided. After all, noise/thresholds/settings may cause the receiver to interpret the analogue signal in a different way. Use a scope to ensure signal integrity, then debug what the receiver reckons it has received.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #31 on: November 29, 2015, 03:34:55 pm »
Sometimes you can compensate or work around sample memory limitations to an extend, but at the end of the day if the sample memory is small then this will limit a scope's performance.

OK, convince me for a "typical case" user of an oscilloscope looking at the time-domain representation of a signal. Oscilloscope != spectrum/vector/network analyser.

Why should I need to "convince you" for a "typical case" (whatever that is). The simple fact is that more sample memory allows you to maintain a higher sample rate at longer time base settings. If that's of any use depends on what you want to do.

Quote
If you have a long duration signal, when isn't is sufficient to digitise at the full rate? Provided you then present three values at each displayed point: mean, min and max so as to attempt to mimic the delightful informative "fuzziness" that can be seen on analogue scopes.

Just to give you an example, I use the large memory of my scope to sample an larger chunk of a non-standard communication signal that I can then pass through specific analysis functionality my scope offers. If the sample memory was smaller then I'd either have to sample a smaller time segment (which would be below the sample size which I require for analysis), or reduce the sample rate which means I'll lower my nyquist frequency (and depending on how small the sample memory was it might well be to an extend where it's no longer useful) and risk aliasing.

Quote
Background: I've been using analogue scopes for 40 years, and digital scopes for 25 years. I have my own opinions as to the advantages, disadvantages, and overlap of each. I like analogue displays, and eye diagrams (and similar "heat maps") when appropriate. I once (in ~1992) had a student build a deep digitiser for just your use case, analysis of comms signals.

That is all fine and well but frankly pretty irrelevant to what mid-range scopes are available in 2015 (which is what this thread is about).

I'd also like to see your student building a digitizer that lets me do the same analysis as my high end scopes  ;)

Quote
I'm a strong believer of debugging digital signals in the digital domain, and that almost all signals are analogue (except where we choose to interpret them as digital).

Well, there aren't really any digital signals, as all physical signals are by nature analog. What is digital is the information we assign to an analog signal with specific parameters, so yes, I agree.

Quote
Hence I will claim that anything equivalent to using a scope to capturing and decoding, for example, the data on an serial peripheral interface (SPI) is misguided. After all, noise/thresholds/settings may cause the receiver to interpret the analogue signal in a different way. Use a scope to ensure signal integrity, then debug what the receiver reckons it has received.

Great, so instead of using one instrument which has the capabilities to thoroughly analyze the signal on the physical and logical level to make sure the input signal is really OK (and just looking on the signal integrity is a far cry from making sure it is!) you want to rely on a receiver which (especially if part of the UUT) might or might not work, and if not you can't really say if its a problem with the sink or the source because you don't know what's actually going in (and it's not just the physical signal where problems can occur). You could of course then try another receiver, and if it works then conclude that the first receiver is broken. But it might just be that there *is* a problem with the source, and the second transmitter just has a wider tolerance in some of its specifications (or a different firmware, or some other minor difference) which allows it to work where the first receiver didn't. Which means while you now believe that your transmission system is now working again there's still a problem, and it will stop working as soon as you replace the sink with one with slightly tighter specs. All while you could have just used the scope to listen to what's going on over the transmission line, which would have shown instantly that there's a flaw in some of the message formats sent by the source.

You stated that you have over 40 years experience with analog scopes, and please don't take this the wrong way but I think that shows. That example above is pretty much taken from reality (a case I witnessed where the EE relied on the receiver to verify if the transmission system was working when it really wasn't; as it turned out he did so because that's the approach he learnt many moons ago and he didn't really know how to work with the advanced features of the scope he had on his bench), and it reflects some of the problems I see pretty often with older engineers that have trained, learnt and spent most of their time on analog scopes and simple DSOs, and who don't really have a grip on how to make use of the advanced analysis capabilities of a modern mid-range or high-end DSO (which at the end of the day are pretty much signal analyzers). It doesn't mean they are bad engineers (a lot of them are actually pretty good) but quite often they waste time and effort working around things that with modern scopes no longer are a problem.

But enough of that.

I'd have used your approach of using the receiver if I was stuck in a time warp in 1999 or if the only scope available was an analog scope (or a primitive DSO), but in this day and age I'd rather use the capabilities of a modern high end DSO to make sure that the input signal is fine, and then focus on the actual area of the problem.

Anyways, that's all not really relevant to this thread.
« Last Edit: November 29, 2015, 03:46:17 pm by Wuerstchenhund »
 

Offline blueskull

  • Supporter
  • ****
  • Posts: 12878
  • Country: cn
  • Power Electronics Guy
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #32 on: November 29, 2015, 10:16:46 pm »
OP,

If you want long recording memory, check out Rigol DS6104, 140Mpts. Warning: loud fan and some minor software bugs, no big issues though.

For fast waveform update, check out MSOX3k/4k. Warning: short memory, but with segmented memory this is not a very serious problem.

For "industry standard" UI and spectrum analyzer: MDO3k/4k. Warning: bad user experience. Generally I never felt Tek scopes easy to use.

For making Wuerstchenhund happy: LeCroy WaveRunner 3k. Warning: not widely used as HPAK or Tek, but specs look nice. I never used one.

-------------------------------- Personal experience --------------------------------

I returned my DS6102 1 week after received it. Good device, but firmware has minor bugs (ETS not working), and its fan is noisy.

I'm now using a MSOX3104A, with no apparent problems encountered yet. Does not have ETS, but I bet you do not need it.

I have access to some 200MHz DPOs and a 2GHz DPO. The 2G one is superb, but the price is not really in hobbyist range.

The 200MHz ones are next to crap. Long boot time, slow update rate, and highly incompatible active probe interface.
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 10694
  • Country: gb
    • Having fun doing more, with less
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #33 on: November 29, 2015, 10:35:42 pm »
You acknowledge that you have a fairly specialised requirement, and I don't disagree with your assessment of the right tool for your job. I suspect I would come to the same conclusion as you.

My question, which you chose not to address, was about tools for less specialised, more general requirements.

Quote
Hence I will claim that anything equivalent to using a scope to capturing and decoding, for example, the data on an serial peripheral interface (SPI) is misguided. After all, noise/thresholds/settings may cause the receiver to interpret the analogue signal in a different way. Use a scope to ensure signal integrity, then debug what the receiver reckons it has received.

Great, so instead of using one instrument which has the capabilities to thoroughly analyze the signal on the physical and logical level to make sure the input signal is really OK (and just looking on the signal integrity is a far cry from making sure it is!) you want to rely on a receiver which (especially if part of the UUT) might or might not work, and if not you can't really say if its a problem with the sink or the source because you don't know what's actually going in (and it's not just the physical signal where problems can occur). You could of course then try another receiver, and if it works then conclude that the first receiver is broken. But it might just be that there *is* a problem with the source, and the second transmitter just has a wider tolerance in some of its specifications (or a different firmware, or some other minor difference) which allows it to work where the first receiver didn't. Which means while you now believe that your transmission system is now working again there's still a problem, and it will stop working as soon as you replace the sink with one with slightly tighter specs. All while you could have just used the scope to listen to what's going on over the transmission line, which would have shown instantly that there's a flaw in some of the message formats sent by the source.

Well of course you have to examine the entire signal chain to determine where the fault is being introduced. That's statin' the bleedin' obvious! N.B. receiver != radio receiver, except in special cases.

I have seen cases where people have relied on the interpretation of a signal by a general purpose oscilloscope, thought they did (or didn't!) see any fault - only to find out that the real receiver was interpreting the signal in subtly different ways. In those cases the "added functionality" actively misled them and delayed their finding the problem.

Quote
You stated that you have over 40 years experience with analog scopes, and please don't take this the wrong way but I think that shows. That example above is pretty much taken from reality (a case I witnessed where the EE relied on the receiver to verify if the transmission system was working when it really wasn't; as it turned out he did so because that's the approach he learnt many moons ago and he didn't really know how to work with the advanced features of the scope he had on his bench), and it reflects some of the problems I see pretty often with older engineers that have trained, learnt and spent most of their time on analog scopes and simple DSOs, and who don't really have a grip on how to make use of the advanced analysis capabilities of a modern mid-range or high-end DSO (which at the end of the day are pretty much signal analyzers). It doesn't mean they are bad engineers (a lot of them are actually pretty good) but quite often they waste time and effort working around things that with modern scopes no longer are a problem.

You have omitted to recognise that I have been using digital scopes for 25 years.

You cannot possibly know that I have been working on novel instrumentation for 30 of the past 40 years, for many things from optical fibres through multiple LANs and wireless LANs to cellular phone systems (with diversions into road surfaces and building materials). Much of that was in HP Labs, including the RF departments of local universities and with the people that designed the front-ends for their most advanced scopes.

So yes, I am very familiar with understanding new tools' strengths, limitations, and how they can be misleading - just like all tools.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 18544
  • Country: nl
    • NCT Developments
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #34 on: November 29, 2015, 11:14:56 pm »
I have seen cases where people have relied on the interpretation of a signal by a general purpose oscilloscope, thought they did (or didn't!) see any fault - only to find out that the real receiver was interpreting the signal in subtly different ways. In those cases the "added functionality" actively misled them and delayed their finding the problem.
Quote
Being able to find the cause of a problem is a talent in itself and does not have much to do with the tools. More sophisticated tools only make is easier/faster to check/verify parts of a circuit but it is still the person operating it which needs to have the brains to work out where the problem is.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 10694
  • Country: gb
    • Having fun doing more, with less
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #35 on: November 30, 2015, 12:53:05 am »
I have seen cases where people have relied on the interpretation of a signal by a general purpose oscilloscope, thought they did (or didn't!) see any fault - only to find out that the real receiver was interpreting the signal in subtly different ways. In those cases the "added functionality" actively misled them and delayed their finding the problem.
Being able to find the cause of a problem is a talent in itself and does not have much to do with the tools. More sophisticated tools only make is easier/faster to check/verify parts of a circuit but it is still the person operating it which needs to have the brains to work out where the problem is.

Yes.

I would add that a more sophisticated/complex tool has more and more subtle ways of leading people astray. That's especially true if they are not already familiar with it and can't/won't take the time to learn what  it is doing and how to use it.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #36 on: November 30, 2015, 12:00:30 pm »
You acknowledge that you have a fairly specialised requirement, and I don't disagree with your assessment of the right tool for your job. I suspect I would come to the same conclusion as you.

Yes, my requirements are certainly special.

Quote
My question, which you chose not to address, was about tools for less specialised, more general requirements.

I haven't ignored it, I just don't know what you mean with "general requirements", bearing in mind that this thread is essentially about choosing a upper mid-range 1GHz scope, not something for say a beginner who dabbles a bit in electronics.

In my experience, people don't invest that kind of money if their requirements aren't somewhat special.

Quote
Well of course you have to examine the entire signal chain to determine where the fault is being introduced. That's statin' the bleedin' obvious!

It may be obvious, but as I said, for the stated example just checking the input signal integrity on a scope isn't sufficient to determine if there's a problem with the input signal or not. And my point was simply that a (somewhat) modern advanced mid-range or high-end scope allows you to check both the physical as well the logical layer to make sure the input signal is OK.

Quote
N.B. receiver != radio receiver, except in special cases.

I didn't say anything about a radio.

Quote
I have seen cases where people have relied on the interpretation of a signal by a general purpose oscilloscope, thought they did (or didn't!) see any fault - only to find out that the real receiver was interpreting the signal in subtly different ways. In those cases the "added functionality" actively misled them and delayed their finding the problem.

Yes, that happens sometimes, but that's pretty independent of the age and the complexity of the test equipment at hand (I've seen people being misled down the path with simple analog scopes or even a voltmeter, and I certainly wasn't immune to getting misled by silly mistakes, too). The important bit is to recognize the error, correct it and learn from it.

But the fact that people can sometimes let themselves misled by stuff is certainly no reason to not use the capabilities of a modern advanced scope.

Quote
You have omitted to recognise that I have been using digital scopes for 25 years.

I recognized that, but while I certainly don't doubt your experience quite frankly it's not saying much. "Digital scopes" can very well mean something along the lines of HP 54600 Series scopes, Tek TDS200, TDS700 Series or any other similar simple scope (and especially Tek and HP didn't even come up with any notable advanced capabilities in their digital scopes until some 15 years ago).

Actually, had you stated something like "worked the last 3 years with a <insert some upper mid-range or high-end scope>" then this would have been way more relevant to the topics at hand.

Quote
You cannot possibly know that I have been working on novel instrumentation for 30 of the past 40 years, for many things from optical fibres through multiple LANs and wireless LANs to cellular phone systems (with diversions into road surfaces and building materials). Much of that was in HP Labs, including the RF departments of local universities and with the people that designed the front-ends for their most advanced scopes.

Sounds like some great experience, but as you said right at the beginning I can't possibly know that, so I have to go from what I read on this forum.

I guess what you don't know about me is that aside from engineering I'm also spending a lot of time in directing various (mostly international) engineering projects, managing/guiding engineers, teaching/training and hiring engineering staff. And one thing I quickly noticed is that a lot of older engineers who spent most of their working life with analog scopes (and tout it as an advantage) have pretty much very little idea how to use a modern advanced scope properly. Not all of them of course, some did adapt, but there's a large number who essentially treats any DSO like an analog scope. We sometimes give them a simple problem to solve (i.e. give them a high end scope and something simple to measure, i.e. a communications link), and you can see when they start fiddling with cursors that they have no idea how to get all the right information out of that scope (and it's not a problem if they just can't operate that specific scope model, they're even allowed to ask i.e. how to enable certain functionality or where they find the setting for XYZ).

It doesn't necessarily mean they won't get hired, though, if they can demonstrate that they are still good engineers and not stuck in the past then they will just receive some additional training. But quite often I see some kind of reluctance to use that "newfangled stuff", and that's the point where it's clear it's not gonna work out. We're working in a very busy and challenging environment, and knowing your basics like how to do some analysis work on a high end scope is pretty much expected.

So forgive me when I'm a bit sceptical when someone touts 40yrs of experience with analog scopes and 25yrs with digital scopes in a thread about purchasing a new upper mid-range scope and believes that this is somehow an advantage, because in my experience more often then not it isn't (again, not doubting your experience of course). The last 15 years have brought large improvements what high end scopes can do, and frankly what happened back in the analog days or even in the early 90's in terms of digital scopes is not of much relevance for that.

Quote
So yes, I am very familiar with understanding new tools' strengths, limitations, and how they can be misleading - just like all tools.

Good for you, and I agree. Especially the latter part can't be stressed enough.
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #37 on: November 30, 2015, 12:06:09 pm »
I would add that a more sophisticated/complex tool has more and more subtle ways of leading people astray. That's especially true if they are not already familiar with it and can't/won't take the time to learn what  it is doing and how to use it.

Any professional should know his tools. Where I work it's pretty much expected that engineers take time and familiarize themselves with new equipment they might work with. That includes having a look at the manual, and taking say a new scope and play around with it for half an hour.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 18544
  • Country: nl
    • NCT Developments
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #38 on: November 30, 2015, 12:17:02 pm »
I would add that a more sophisticated/complex tool has more and more subtle ways of leading people astray. That's especially true if they are not already familiar with it and can't/won't take the time to learn what  it is doing and how to use it.
Any professional should know his tools. Where I work it's pretty much expected that engineers take time and familiarize themselves with new equipment they might work with. That includes having a look at the manual, and taking say a new scope and play around with it for half an hour.
I agree. If you get fooled by your tools then you need to learn how to use them first. When doing complicated measurements (trigger setups) I do some tests first to verify I got the test setup right so I know the results are meaningful.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #39 on: November 30, 2015, 04:13:07 pm »
If you want long recording memory,

Plus long nights searching through all that memory on a dog slow scope to find the event you need, because like all Rigol scopes the DS6000 has no sane search facility whatsoever, aside from some simple mask testing facility.  :--

The DSO6000 also not even anywhere close to be in the same performance ballpark as the other scopes discussed in this thread, as its essentially just a blown up DS4000 (which itself isn't exactly a shining beacon of a great entry level scope), plus there are the typical firmware issues, and the questionable Rigol support.

Why anyone would spend that much money on an inferior Chinese B-brand is beyond me.

Quote
check out Rigol DS6104, 140Mpts. Warning: loud fan and some minor software bugs, no big issues though.

I'm not sure I'd call the fact that almost 5 years after the scope came out a major part of the functionality (ETS) still doesn't work a "minor software bug".   :palm:

But I guess the issues were still big enough for you to return your DS6102, didn't you?  >:D

Quote
For "industry standard" UI and spectrum analyzer: MDO3k/4k

Not sure why you consider Tek's archaic UI "industry standard", as it's neither really a standard and even less so an "industry standard". A better word to describe it might be "retro", and even that would be overly generous. Most other scope UIs have developed a lot since the 90's. Tek obviously hasn't, but then that's true in other areas as well.

Quote
For making Wuerstchenhund happy: LeCroy WaveRunner 3k.

Frankly, it would have made me more happy if you had at least looked at the actual requirements first, which were for a 1GHz 5Gsa/s scope where all contenders were pretty much in the >$14k range (so I'm not sure why you even list entry level scopes). Considering that the WaveSurfer 3000 is a 4GSa/s scope with a max bandwidth of 750MHz it's not even a contender.

Quote
Warning: not widely used as HPAK or Tek

You might be surprised, especially when considering that the WS3k only came out in Q2/2014. It's eating very well into Keysight's DSOX3000 Series, and is pretty much one of the reasons why Keysight came up with the "new" DSOX3kT in the first place.

Anyways, that hasn't much to do with the topic at hand.
« Last Edit: November 30, 2015, 04:16:11 pm by Wuerstchenhund »
 

Offline blueskull

  • Supporter
  • ****
  • Posts: 12878
  • Country: cn
  • Power Electronics Guy
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #40 on: November 30, 2015, 04:25:27 pm »
W,

The reason I say Tek is "std" is because in so many academic papers scope screen shots are Tek's. I never say I like it, and being said, I don't like Tek.
Sorry for suggesting the wrong product, because I have no idea about LeCroy's line up, and I've never seen one in real life. In my place it is pretty much all Tek and HPAK.
Rigol has segmented memory and advanced triggers, just not as good as other brands'. If you think thoroughly about the signal to be tested, and tailor a proper trigger condition for it, it will work just fine.

BTW, ETS is not the biggest reason I returned that unit. The reason is I found a $6000 MSOX3104A offer and the carelessness of Rigol quality control (angled BNC connector) and really loud fan.
I managed to measure phase delay with other methods, as usual, improvised tool works the best on its intended job.
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #41 on: November 30, 2015, 07:03:37 pm »
The reason I say Tek is "std" is because in so many academic papers scope screen shots are Tek's.

That may well be (although that probably depends on the type of academic papers, i.e. science/physics its pretty much LeCroy due to their history there) but like in many other things education doesn't necessarily reflect what you find in industry.

Quote
Sorry for suggesting the wrong product, because I have no idea about LeCroy's line up, and I've never seen one in real life. In my place it is pretty much all Tek and HPAK.

Well, as I said, it's a bit strange to find recommendation for what essentially are entry-level scopes in a thread like this.

Quote
Rigol has segmented memory and advanced triggers, just not as good as other brands'. If you think thoroughly about the signal to be tested, and tailor a proper trigger condition for it, it will work just fine.

Yes, but segmented memory doesn't help when the search functionality is crap or better semi-nonexistent. And its not just memory search and triggers that are not as good as any other scope in that price range, there's also FFT which does what, 4k points or so?

Quote
BTW, ETS is not the biggest reason I returned that unit. The reason is I found a $6000 MSOX3104A offer and the carelessness of Rigol quality control (angled BNC connector) and really loud fan.

Well, it's Rigol. As I said, why would anyone in their right mind spend that amount of money on an inferior Rigol scope when the same money could buy you a proper scope?

Quote
I managed to measure phase delay with other methods, as usual, improvised tool works the best on its intended job.

That's true.
 

Offline H.O

  • Frequent Contributor
  • **
  • Posts: 636
  • Country: se
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #42 on: December 01, 2015, 06:24:31 am »
Quote
And its not just memory search and triggers that are not as good as any other scope in that price range, there's also FFT which does what, 4k points or so?

It's actually 2048 points according to the following app note: FFT data depth. Though, this app note says it's 700 for the 4k and 6k but in the brief FOR that app note it again says 2048:
Quote
What data set size is used to calculate the FFT on the DS series of scopes?
The FFT is calculated using the displayed data set. DS1000E/D/B/CA series scopes = 600 Data points DS1000Z Series (Displayed) = 1200 (Memory) = 16,384 DS2000 Series = 2048 points DS4000 series = 2048 points DS6000 series = 2048 points
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #43 on: December 01, 2015, 10:27:09 am »
It's actually 2048 points according to the following app note: FFT data depth. Though, this app note says it's 700 for the 4k and 6k but in the brief FOR that app note it again says 2048:
Quote
What data set size is used to calculate the FFT on the DS series of scopes?
The FFT is calculated using the displayed data set. DS1000E/D/B/CA series scopes = 600 Data points DS1000Z Series (Displayed) = 1200 (Memory) = 16,384 DS2000 Series = 2048 points DS4000 series = 2048 points DS6000 series = 2048 points

Wow, that's even worse than I thought! 2kpts FFT in a scope Rigol wants something in the region of $6k for?  |O

Anyways, it just shows that Rigol's DS6000 is pretty much a toy compared to what the big brands give you in that price class.
 

Offline stejo780

  • Contributor
  • Posts: 12
  • Country: no
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #44 on: December 28, 2015, 05:53:46 pm »
Wow, the discussion did really go warm! Thanks again for all input.

I ended up with a R&S RTE1204, the extra bandwidth did not cost much extra and we did get a pretty good deal trading in an old scope and bought demo used probes. LeCroy did not have any opportunity to come and do a demo (it was very short notice...) and I did not want to order without first trying them out. Still think they look interesting, so will include them on the shortlist next time  :). The offer from LeCroy was also more significantly more expensive as mentioned above.

The scope has arrived now, but I'm away from work, so it's not yet unboxed.

Cheers,
Stefan
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 2804
  • Country: gb
  • Only sparsely active but available via PM
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #45 on: December 28, 2015, 06:19:42 pm »
I ended up with a R&S RTE1204, the extra bandwidth did not cost much extra and we did get a pretty good deal trading in an old scope and bought demo used probes.

Well done!

Quote
LeCroy did not have any opportunity to come and do a demo (it was very short notice...) and I did not want to order without first trying them out.

A wise decision, no matter which manufacturer.

Quote
The scope has arrived now, but I'm away from work, so it's not yet unboxed.

When you have used it for a while it'd be great if you could let us know your impressions and what you like and not. Even better would be a review  ;)
 

Offline stejo780

  • Contributor
  • Posts: 12
  • Country: no
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #46 on: December 29, 2015, 04:51:26 pm »
I don't think I'll be able to give a full review, but I will definitely share my impressions when i have more experience with it.

So far, I can say that it's absolutely a huge upgrade to what I used to have on my bench. The LED intensity, which was mentioned earlier in the thread, can easily be adjusted, but the setting did not persist when the scope is rebooted. I will have to look in to presets to solve that, it was somewhat annoying since the LEDs are really intense in the original setting.

Another thing that I wanted in one occasion was to look at the quote between two RMS measurements. This is not possible according to the application engineer, but he will ask the development to look in to it. Should be a rather simple thing to implement one would imagine.

Cheers,
Stefan



 

Offline blueskull

  • Supporter
  • ****
  • Posts: 12878
  • Country: cn
  • Power Electronics Guy
Re: 5 Gsample/s, 1 GHz, 4 ch scope advise
« Reply #47 on: December 29, 2015, 05:02:32 pm »
It just popped up in my mind, would you want to try some PC solution? Say, a PicoScope 6404C? A 500MHz, 5Gsps scope with remarkably low noise. 170kwfms/s and HUGE sampling memory.

The lowest quote I've ever got is $5.5k, so not too expensive. Reports say even a 350MHz PicoScope can go up to 1GHz, just with higher attenuation. The 500MHz one should be much better, since its front end circuit is completely re-designed.

If high trigger rate and huge memory along with USB connectivity is your concern, consider it. It has built in ETS as well, and an even faster trigger mode for smaller sampling memory. Also, it comes with segmented memory, serial deciding and many other goodies for free, along with some very good quality probes.

The Chip Whisperer guy has a review video of its 350MHz little brother.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf